CASE STUDY

At the time of the analysis, Debenhams was pleased to find that its AOL portal deal was highly
profitable. While only 0.62% of all visits came through AOL, the conversion rate of 2.6% easily
outranked the other portals. More importantly, the average purchase by an AOL member was more
than twice and almost three times the purchase amount of the two other closest-competing portal
deals. In addition to the referrer performance metrics, Blue Martini also provided Debenhams with a
detailed list of search keywords that people used at these major Internet portals to get to Debenhams,
enabling the retailer to plan out strategies for paid ads based on keywords.
T HE PERFECT MATCH: EV ALUATING THE EFFECTIVENESS OF ONLINE CROSS
SELLING
Retailers rely on teams of marketers to recommend cross-sell and up-sell products.These days with rapidly
changing purchasing trends and inventory lists stretching into thousands of products, it is not always
feasible for marketers to keep the cross-sell and up-sell lists up to date. An automatic product recommender
model based on association rules can be used effectively to complement the recommendations
provided by the marketing experts.
Figure: Product affinities based on purchase patterns at Debenhams can help marketers make cross-sell recommendations.
The association algorithm in the Blue Martini Business Intelligence suite identified over 200
significant product associations. For example, the Debenhams Web site data shows that 41% of all
people (confidence) who bought fully-reversible bath mats bought Egyptian cotton towels and that a
customer who bought a bath mat was 456 times more likely (lift) to buy an Egyptian cotton towel
than a random visitor. Yet the site recommends another towel set—with only 1.4% of all people who
purchased the fully reversible bath mats selecting this option. Similarly, the data indicates that 25% of
the purchasers of the white T-shirt bra bought the plunge T-shirt bra and were 246 times more likely
than average to do so. Again the site’s suggestion did not perform as well—only 1% of the white bra
purchasers opted for the black underwire bra. Blue Martini recommended that Debenhams share these
associations with merchandisers and implement an automated product recommender.
U NDERSTANDING THE GEOGRAPHICAL DISTRIBUTION OF CUSTOMERS
On the whole, the Debenhams online profile indicates a successful e-business channel.The site
has over two million page views per week and grows by over 6,000 new customers per week.
Not surprisingly, UK customers generate most of Debenhams’ revenue with major urban areas
such as London, Manchester, Bristol, and Glasgow emerging as major revenue sources.And, even
though Debenhams only fulfills delivery requests within the UK, there are a measurable number
of customers from the US and Australia. The customers from outside the UK are indicators of
the on-going success of online wedding list—which accounts for over half of non-UK sales.
Blue Martini Software Customer Case Study - 6
DEBENHAMS